https://uberduck.ai/ logo
Join Discord
Powered by
# machine-learning
  • m

    mega b

    05/18/2021, 3:58 AM
    😐
  • m

    mega b

    05/18/2021, 3:58 AM
    ?warn @User Don't make the models say anything offensive
  • m

    mega b

    05/18/2021, 3:59 AM
    dont make it say that 😐
  • m

    mega b

    05/18/2021, 4:00 AM
    its the first rule
  • m

    mega b

    05/18/2021, 4:01 AM
    ?warn @User idk what to say, lol
  • m

    mega b

    05/18/2021, 4:01 AM
    read the damn rules
  • m

    mega b

    05/18/2021, 4:01 AM
    we simplified it as much as possible
  • m

    mega b

    05/18/2021, 4:02 AM
    You can make it say that
  • m

    mega b

    05/18/2021, 4:03 AM
    just nothing that harasses a person / group / community
  • m

    mega b

    05/18/2021, 4:03 AM
    no
  • m

    mega b

    05/18/2021, 4:04 AM
    best not to share it with anyone at all
  • m

    mega b

    05/18/2021, 4:05 AM
    keep that info to your self
  • m

    mega b

    05/18/2021, 4:06 AM
    ¯\_(ツ)_/¯
  • m

    mega b

    05/18/2021, 4:06 AM
    just follow the rules, alright
  • m

    mega b

    05/18/2021, 4:06 AM
    I cant really change your mind into doing so, but consider it
  • a

    Alexius08

    05/18/2021, 10:21 AM
    Current alignment chart for ongoing training of the Gordon Ramsay TTS model.
  • d

    deathGripsfanforever

    05/18/2021, 10:26 AM
    I love Gordon Ramsay
  • a

    Alexius08

    05/18/2021, 2:04 PM
    Reached up to epoch 180 before validation loss started rising unusually fast. Had four epochs with good alignment charts and validation loss of less than 0.2.
  • a

    Alexius08

    05/18/2021, 2:13 PM
    Epoch 173
  • a

    Alexius08

    05/18/2021, 2:13 PM
    Epoch 180
  • a

    Alexius08

    05/18/2021, 2:13 PM
    Which one has the better alignment chart?
  • a

    Alexius08

    05/18/2021, 3:50 PM
    Snippet from the Navy Seal Copypasta using the Epoch 180 model.
  • a

    Alexius08

    05/18/2021, 3:50 PM
    Snippet from the Navy Seal Copypasta using the Epoch 173 model.
  • a

    Ask for dataset = EATEOT

    05/18/2021, 3:57 PM
    Gordon is trying to communicate with us through the static
  • p

    pi

    05/18/2021, 4:20 PM
    your dataset is probably not large enough/clean enough
  • b

    brilliam.de

    05/18/2021, 4:28 PM
    uh oh
  • a

    Ask for dataset = EATEOT

    05/18/2021, 4:29 PM
    flatline
  • h

    HoodyP

    05/18/2021, 4:40 PM
    Well since it’s yellow and in line, it means that its mostly progressing! What you dont want was the yellow line disappearing or being scattered all around the place instead of a line
  • h

    HoodyP

    05/18/2021, 4:41 PM
    This is what you dont want, but don’t worry, it will fix up over time by more training!
  • b

    brilliam.de

    05/18/2021, 4:43 PM
    okay awesomeA!
1...959697...1068Latest